Adding worst model + some comments

src/main.py

@@ -188,18 +188,26 @@ def allModels(df):
         arrayColumns = [x[i]]
         for j in range(i+1,len(x)):
             xValues = dfTemp[arrayColumns]
-            for k in range(0,len(modelArray)):
-                if modelArray[k] == "KNN":
-                    model = model_switch(1)
-                elif modelArray[k] == "Classifier":
-                    model = model_switch(2)
-                else:
-                    model = model_switch(1)
-                print("Model used : ",modelArray[k], "---- Case : ",model)
-                print("X values used : ",arrayColumns)
-                accu = customTrainingRaw(model,xValues,y,3)
-                it = [modelArray[k],arrayColumns,accu]
-                datas.append(it)
+            
+            # Knn model train
+            model = model_switch(1)
+            accuKnn = customTrainingRaw(model,xValues,y,3)
+            print("Model used : Knn ---- Case : ",model)
+            print("X values used : ",arrayColumns)
+
+            # Tree model train
+            model = model_switch(3)
+            accuTree = customTrainingRaw(model,xValues,y,3)
+            print("Model used : Tree ---- Case : ",model)
+            print("X values used : ",arrayColumns)
+
+            
+            dico = dict()
+            setUp = [arrayColumns.copy(),dico]
+            setUp[1]['Knn'] = accuKnn
+            setUp[1]['Tree'] = accuTree
+            datas.append(setUp.copy())
+            
             arrayColumns.append(x[j])
     return datas
 
@@ -216,25 +224,61 @@ def customTrainingRaw(model, x, y,res=-1):
     print(accuracy_score(ytest, ypredit))
     return accuracy_score(ytest, ypredit)
 
-def bestModelFinder(datas):
-    maxi = 0
-    knnMean= 0
-    treeMean= 0
+def showStat(datas):
+    fig, ax = plt.subplots()
+    x_data = []
+    y_dataKnn = []
+    y_dataTree = []
     for i in range(0,len(datas)):
-        if datas[i][0] == 'KNN':
-            knnMean += datas[i][2]
-        else:
-            treeMean += datas[i][2]
-        if (datas[i][2] > maxi):
-            maxi = datas[i][2]
+        x_data.append("/".join(datas[i][0]))
+        y_dataKnn.append(datas[i][1]['Knn'])
+        y_dataTree.append(datas[i][1]['Tree'])
+
+    ax.scatter(x_data, y_dataKnn, label=f'Y = Knn')
+    ax.scatter(x_data, y_dataTree, label=f'Y = Tree')
+    ax.set_xlabel('Axe X')
+    ax.set_ylabel('Axe Y')
+    ax.legend()
+    plt.show()
+
+
+def bestModel(datas):
+    max = 0
+    min = 1
+    for i in range(0,len(datas)):
+        if(datas[i][1]['Knn'] < min):
+            min = datas[i][1]['Knn']
+            resMin = datas[i]
+            modelMin = 'Knn'
+        elif datas[i][1]['Tree'] < min:
+            min = datas[i][1]['Tree']
+            resMin = datas[i]
+            modelMin = 'Tree'
+
+        if(datas[i][1]['Knn'] > max):
+            max = datas[i][1]['Knn']
             res = datas[i]
-    print("BEST CHOICE IS :", res)
-    print("Knn mean accuracy_score : ", mean(knnMean))
-    print("Knn variance accuracy_score : ", variance(knnMean))
-    print("Knn ecart-type accuracy_score : ", stdev(knnMean))
-    print("Tree mean accuracy_score : ", mean(treeMean))
-    print("Tree variance accuracy_score : ", variance(treeMean))
-    print("Tree ecart-type accuracy_score : ", stdev(treeMean))
+            model = 'Knn'
+        elif datas[i][1]['Tree'] > max:
+            max = datas[i][1]['Tree']
+            res = datas[i]
+            model = 'Tree'
+    print("Best model : ",model," columns : ",res[0]," Accuracy : ", res[1][model])
+    print("Worst model : ",modelMin," columns : ",resMin[0]," Accuracy : ", resMin[1][model])
+
+df = read_dataset('data.csv')
+
+# Affiche la répartitions des objets stélaires dans la base de données
+#showData(df)
+
+# Affiche le meilleur models avec les meilleurs colonnes entre KNeighborsClassifier et DecisionTreeClassifier
+#datas = allModels(df)
+#bestModel(datas)
 
+# Génère un nuage de points affichant l'accuracy du model Knn et TreeClassifier en fonction des colonnes utilisées.
+datas = allModels(df)
+showStat(datas)
+bestModel(datas)
 
-main()
+# Affiche un menu permettant de choisir le model à entrainer, ainsi que des stats suplémentaires
+# main()